First page Back Continue Last page Summary Graphics
Here the talk ends and I took questions.
The guy who gave the S/390 port talk just before mine asked about the performance impact of running things inside a virtual machine. Conveniently, I had no hard numbers handy.
There was a suggestion about an alternate means of intercepting system calls which didn't seem all that different from ptrace.
I had to clear up a misconception about UML not really doing its own scheduling. I emphasized that UML is the full-blown kernel with the scheduler and VM system and everything else a self-respecting kernel needs.